Abstract:By comparing and analyzing the status quo of and the hidden cognitive principles in Machine Translation and Translation Memory, this paper concludes that CAT software is a great achievement by using a computer to describe formal features in interlingual transmission as semantic barriers remain unsettled in translating practice. Although Machine Translation and Translation Memory take different attitudes towards semantic barriers, the former tries to directly analyze and process current translation itself, the latter conducts formal computation on interlingual transmission by using previous cumulated experience, however, they both embody common human cognitive economic trend on the basis of cognitive strategy, therefore, when CAT software is designed, the method should be innovated and Machine Translation and Translation Memory should be simultaneously integrated to let the two support each other so that better translationaided effect can be obtained.